Skip to main content

m02 · Volume conduction and Poisson’s equation

Warning

Education only. Not for diagnosis, treatment or any clinical decision.

In 60 seconds

The heart sits in a volume conductor: blood, tissue and the torso carry the current that membrane sources inject. At the frequencies of bioelectric signals the fields are quasi-static: current is conserved instant by instant, and the potential obeys Poisson’s equation, \(\nabla\cdot(\sigma\nabla\phi) = -I_v\). In a uniform infinite medium its solution is the point-source formula of m01. Real anatomy is not uniform, so this module adds what changes at a boundary: potential and normal current are continuous, and for a flat interface the method of images gives the exact answer. An insulating boundary doubles the potential; a better-conducting medium on the electrode’s side lowers it. For any geometry, a grid solves the problem, and where the interface falls on the grid decides how accurate it is. Finally, reciprocity turns the question around: the field you get by injecting current at an electrode is that electrode’s sensitivity to every source, its lead field. A bipole’s lead field falls off faster with depth, which is why it rejects far-field signals.

From the field

  • The medium around the electrode matters. Brody analyzed how the conductive blood mass inside the heart changes the relation between cardiac sources and surface leads (Brody 1956); Rudy and Plonsey studied the role of geometry and inhomogeneities with eccentric spheres (Rudy and Plonsey 1979). The flat-interface model (Equation 3) is the simplest version of that effect.
  • Quasi-static is an assumption worth stating. Electrophysiological fields can be treated as quasi-static (Plonsey and Heppner 1967): capacitive, inductive and propagation effects in the volume conductor are neglected, so the potential everywhere follows the sources instantly (Equation 1).
  • Any electrode records a weighted sum of all sources. Reciprocity says the weights are the electrode’s lead field (Equation 5). Close electrode pairs weight nearby tissue more, which is the far-field rejection of m03 seen from the electrode’s side.
  • Numerical models carry their own error. On a grid, putting a blood–tissue interface on a row of nodes instead of between rows turns a second-order method into a first-order one (lab part C). Forward models for ECG imaging (m32) and digital twins (m33) depend on that accuracy.

Derivations

Symbols used throughout. \(\phi\) potential. \(\sigma\) conductivity (S/m); \(\sigma_1\), \(\sigma_2\) on the two sides of an interface. \(\mathbf J\) current density. \(I_v\) current source density, current injected per unit volume (positive for a source). \(I\) point-source current. \(k\) reflection coefficient. \(h\) grid spacing (mm). \(\psi\) potential produced by unit current injected at an electrode. \(\mathbf p\) current dipole moment. Relative units, distances in mm.

Poisson’s equation

In the extracellular medium, current density is ohmic, \(\mathbf J = -\sigma\nabla\phi\). Treating the fields as quasi-static (Plonsey and Heppner 1967), charge does not accumulate, so the current leaving any small volume equals what sources inject into it, \(\nabla\cdot\mathbf J = I_v\):

\[ \nabla\cdot\left(\sigma\,\nabla\phi\right) = -I_v \tag{1}\]

In a uniform medium this is \(\nabla^2\phi = -I_v/\sigma\), whose solution for a point source is the m01 formula (eq-m01-point). Away from sources, \(\phi\) satisfies Laplace’s equation. A 2-D Gaussian source of total current \(I\) and width \(s\) has the exact potential \(-\frac{I}{2\pi\sigma}\left(\ln r + \tfrac12 E_1\!\left(r^2/2s^2\right)\right)\), where \(E_1\) is the exponential integral; the lab uses it to check a numerical solver.

Key takeaways

  • Poisson’s equation is current conservation plus Ohm’s law.
  • Changes in conductivity enter through \(\nabla\cdot(\sigma\nabla\phi)\), not only through the sources.

Interfaces and images

Where two media meet, the potential is continuous and so is the current crossing the interface:

\[ \phi_1 = \phi_2, \qquad \sigma_1\,\frac{\partial\phi_1}{\partial n} = \sigma_2\,\frac{\partial\phi_2}{\partial n} \tag{2}\]

For a point source \(I\) in medium 1 at depth \(d\) below a flat interface, both conditions are met exactly by the method of images. Medium 1 sees the source plus a mirror source of strength \(kI\) at height \(+d\); medium 2 sees the source alone, with strength \((1 + k)I\):

\[ k = \frac{\sigma_1 - \sigma_2}{\sigma_1 + \sigma_2}, \qquad \phi_1 = \frac{I}{4\pi\sigma_1}\left(\frac1r + \frac{k}{r'}\right), \qquad \phi_2 = \frac{I}{2\pi(\sigma_1 + \sigma_2)\,r} \tag{3}\]

where \(r\) and \(r'\) are distances to the source and its image. Three cases: equal media (\(k = 0\)) give the m01 point source; an insulating medium 2 (\(k = 1\)) doubles the potential at the boundary; a better-conducting medium 2 shunts current and lowers the potential in it by the factor \(2\sigma_1/(\sigma_1 + \sigma_2)\) compared with medium 1 everywhere. The same \(k\) applies to line sources in two dimensions, with \(\ln r\) in place of \(1/r\).

Exact

Assumptions

  • Model: Equation 3 for a unit point source below a flat infinite interface; quasi-static fields. Tested against epml.volume.
  • Parameters: conductivity of medium 1 equal to 1; the ratio \(\sigma_2/\sigma_1\) from 0.01 to 10 and the source depth from 0.5 to 5 mm are free choices, not tissue values.
  • What it ignores: curved and finite interfaces, more than two media, anisotropy, and the electrode itself.

Key takeaways

  • An interface acts like a mirror source of strength \(k = (\sigma_1 - \sigma_2)/(\sigma_1 + \sigma_2)\).
  • An insulating boundary doubles the potential; a better conductor on the electrode’s side lowers it.

Solving on a grid

For any geometry, discretize Equation 1. On a square grid of spacing \(h\), balance the current through the four faces of each node’s cell, with face conductivities taken as harmonic means of the nodes on either side:

\[ \frac{1}{h^2}\sum_{\text{4 neighbors } j} \sigma_{ij}\left(\phi_j - \phi_i\right) = -I_{v,i}, \qquad \sigma_{ij} = \frac{2\sigma_i\sigma_j}{\sigma_i + \sigma_j} \tag{4}\]

The harmonic mean makes the scheme conserve current across a jump in conductivity. For a smooth problem the error falls as \(h^2\). With an interface, the scheme stays second-order only if the interface lies on cell faces, halfway between rows of nodes; on a row of nodes it is shifted by half a cell and the error falls only as \(h\) (lab part C).

Simulation

Assumptions

  • Model: Equation 4 on a 20 × 20 mm square, 20–100 nodes per side, a Gaussian source of width 0.3 mm 2 mm below a flat interface; boundary values from the exact image solution; solved by successive over-relaxation (factor 1.9, tolerance \(10^{-10}\)). The JS solver is tested against a direct sparse solve in epml.volume.
  • Parameters: conductivity 1 below the interface; the ratio above is a free choice.
  • What it ignores: three dimensions, anisotropy, and boundaries other than the fixed outer ring.

Key takeaways

  • A flux-conserving grid with harmonic-mean faces is second-order accurate.
  • Where the interface sits in the grid decides the accuracy.

Reciprocity and lead fields

Inject unit current at an electrode (and remove it far away) and call the resulting potential \(\psi\). By Helmholtz reciprocity, a dipole \(\mathbf p\) at \(\mathbf r_0\) produces at that electrode the potential

\[ V = \mathbf p\cdot\nabla\psi(\mathbf r_0) \tag{5}\]

so \(\nabla\psi\), the electrode’s lead field, is its sensitivity to sources everywhere, whatever the media in between. One solve per electrode gives the sensitivity to every source. In a uniform medium, \(\psi = 1/(4\pi\sigma\lvert\mathbf r - \mathbf r_e\rvert)\) and Equation 5 reduces to the m01 dipole (eq-m01-dipole). For a bipole, \(\psi\) comes from injecting current at one electrode and removing it at the other, so its lead field falls off faster with depth: with 2-mm spacing 1 mm above the tissue, sensitivity at 8 mm depth is 0.5% of that at 0.2 mm, against 3% for a single electrode (lab part D).

Key takeaways

  • An electrode’s sensitivity map is the field it would create if it injected current.
  • Bipoles see shallower tissue than unipolar electrodes.

Clinical correlate

NotePending clinical review

This section describes published findings only. It is not guidance for patient care, and it has not yet been reviewed by an EP physician.

Where this shows up in EP. Every electrogram and every ECG lead is a potential in a volume conductor. The conductive blood inside the heart changes how cardiac sources appear in surface leads (Brody 1956), and geometry and inhomogeneities matter for the ECG (Rudy and Plonsey 1979). ECG imaging (m32) solves the torso problem backwards and needs an accurate forward model; electroanatomic mapping relies on electrodes that sit in blood next to tissue.

What the evidence is. Theoretical analyses (Brody 1956; Rudy and Plonsey 1979; Plonsey and Heppner 1967) and textbook physics (Plonsey and Barr 2007). Their abstracts are not available on PubMed, so this module cites them for their subject, not for numbers.

What a reader should ask:

  • Which media lie between the source and the electrode? Each interface changes the potential (Equation 3).
  • What is the electrode’s lead field? It sets which tissue the signal comes from (Equation 5).
  • How was a model solved? Grid placement and spacing set its accuracy (Equation 4).

Visual summary

Schematic

medium 2 (σ₂)medium 1 (σ₁) images: k = (σ₁ − σ₂)/(σ₁ + σ₂) grid: current balance per cell reciprocity: lead field = ∇ψ
Figure 1: Three tools for real geometry: images for a flat interface, a flux-conserving grid for anything else, and reciprocity to read an electrode’s sensitivity map.

Assumptions

How experts think

Before trusting a forward model or explaining a strange electrogram, ask:

  1. Is quasi-static appropriate? For bioelectric signals it usually is (Plonsey and Heppner 1967).
  2. What are the media and their interfaces? Each one acts like a mirror source (Equation 3).
  3. Is the solution exact or numerical? If numerical, how does the error fall with grid spacing (Equation 4)?
  4. What does the electrode see? Compute or reason about its lead field (Equation 5).
  5. What would change if the conductivity ratio were different? Treat it as uncertain unless it is measured.

One sentence to keep: the potential is current conservation solved in the real geometry, and the electrode’s lead field tells you which tissue it reports.

Cheat sheet

Quantity Result
Poisson’s equation (Equation 1) \(\nabla\cdot(\sigma\nabla\phi) = -I_v\); Laplace’s equation away from sources
Interface conditions (Equation 2) \(\phi\) and \(\sigma\,\partial\phi/\partial n\) continuous
Images (Equation 3) \(k = (\sigma_1 - \sigma_2)/(\sigma_1 + \sigma_2)\); \(\phi_2 = I/(2\pi(\sigma_1 + \sigma_2)r)\)
Special cases Insulator: potential doubled at the boundary; equal media: m01 point source
Finite differences (Equation 4) Harmonic-mean faces; second order if interfaces lie on faces, first order if on nodes
Reciprocity (Equation 5) \(V = \mathbf p\cdot\nabla\psi(\mathbf r_0)\); \(\nabla\psi\) is the lead field

Values used in this module

Value Source
All conductivities: relative, with ratios as free parameters our own code
Bipole lead field at 8 mm against 0.2 mm (2-mm spacing, 1 mm above tissue): 0.5%, against 3% unipolar our own code

Interview questions

Try each question before you open its model answer. In a real interview, say which assumptions you are making.

Q1. Derive Poisson’s equation for a volume conductor.

Model answer

Ohm’s law, \(\mathbf J = -\sigma\nabla\phi\), and current conservation, \(\nabla\cdot\mathbf J = I_v\), give \(\nabla\cdot(\sigma\nabla\phi) = -I_v\) (Equation 1).

Q2. What does “quasi-static” mean here, and why is it acceptable?

Model answer

The potential is taken to follow the sources instantly, neglecting capacitive, inductive and propagation effects in the volume conductor; for electrophysiological signals this is a standard approximation (Plonsey and Heppner 1967).

Q3. What are the conditions at an interface between two conductors?

Model answer

Potential is continuous and the normal current \(\sigma\,\partial\phi/\partial n\) is continuous (Equation 2).

Q4. Explain the method of images for a flat interface.

Model answer

In the source’s medium, add a mirror source of strength \(kI\), \(k = (\sigma_1 - \sigma_2)/(\sigma_1 + \sigma_2)\); in the other medium, use the source alone with strength \((1 + k)I\). Both interface conditions are then met exactly (Equation 3).

Q5. What happens to the potential at an insulating boundary?

Model answer

It doubles: \(k = 1\), so the image adds an equal source at the mirror point, and at the boundary \(r = r'\) (Equation 3).

Q6. An electrode sits in a medium that conducts better than the tissue holding the source. Higher or lower potential than if everything were tissue?

Model answer

Lower, by \(2\sigma_1/(\sigma_1 + \sigma_2)\): the better conductor shunts current (Equation 3; demo m02-d1).

Q7. Why use harmonic means for face conductivities in a finite-difference scheme?

Model answer

Two resistors in series: the face conductance between nodes of different conductivity is the harmonic mean, which conserves current across a jump (Equation 4).

Q8. Your solver converges at first order instead of second. What might be wrong?

Model answer

An interface lying on a row of nodes instead of between rows shifts it by half a cell (lab part C), or the source or boundary treatment is first-order (Equation 4).

Q9. State reciprocity for an electrode and a dipole.

Model answer

\(V = \mathbf p\cdot\nabla\psi(\mathbf r_0)\), where \(\psi\) is the potential from unit current injected at the electrode (Equation 5).

Q10. Why is reciprocity useful in practice?

Model answer

One forward solve per electrode gives its sensitivity to every source location, instead of one solve per source (Equation 5).

Q11. Why does a bipole see shallower tissue than a unipolar electrode?

Model answer

Its lead field is the difference of two nearby electrodes’ lead fields, a dipole-like field that falls off faster with distance (Equation 5; lab part D).

Q12. How would you check a forward solver?

Model answer

Against exact solutions (a Gaussian source, images for a flat interface), by measuring the convergence order as the grid is refined, and by checking current conservation and reciprocity (Equation 4, Equation 3).

Code lab

The lab solves Poisson’s equation exactly and on a grid, adds an interface, and computes lead fields.

Part What you do Data
A Poisson’s equation for a Gaussian source; finite-difference convergence (Equation 1, Equation 4) synthetic
B A flat interface: images, continuity and the electrode-potential factor (Equation 2, Equation 3) synthetic
C Two media on a grid: interface on faces or on nodes (Equation 4) synthetic
D Reciprocity and lead fields of a single electrode and a bipole (Equation 5) synthetic

Each part ends with checks: lines that pass with the code as given. If you change something and a check fails, that is information, not an error to hide.

Self-check

Short checks you can do in your head. Open each answer after you have tried it.

S1. In a uniform medium, what does Poisson’s equation reduce to away from sources?

Answer

Laplace’s equation, \(\nabla^2\phi = 0\) (Equation 1).

S2. σ₁ = σ₂. What is k?

Answer

Zero (Equation 3).

S3. Medium 2 is an insulator. What is k?

Answer

One (Equation 3).

S4. Medium 2 conducts three times better than medium 1. The potential in medium 2, relative to medium 1 everywhere?

Answer

\(2/(1 + 3) = 0.5\) (Equation 3).

S5. You halve the grid spacing of a second-order scheme. The error?

Answer

About a quarter (Equation 4).

S6. What is an electrode’s lead field?

Answer

The gradient of the potential it would create by injecting unit current: its sensitivity to dipoles everywhere (Equation 5).

Literature figures

These figures are worth looking at alongside this module. We link to them instead of copying them: some licenses do not allow reuse here, and others we could not confirm. Each entry says what to look for and where this module explains it.

  • The blood mass and the heart-lead relationship. Brody’s analysis of how intracavitary blood changes the leads (Brody 1956). This module: interfaces and images.
  • Eccentric spheres. Rudy and Plonsey’s model of the heart inside the torso, with geometry and inhomogeneities (Rudy and Plonsey 1979). This module: interfaces, and m32 for the inverse problem.

Further reading

  • Bioelectricity. Plonsey and Barr’s textbook on volume conductors and lead fields (Plonsey and Barr 2007).
  • Quasi-stationarity. Plonsey and Heppner (Plonsey and Heppner 1967).
  • Inhomogeneities and the ECG. Brody (Brody 1956); Rudy and Plonsey (Rudy and Plonsey 1979).
  • Related modules. m01 (sources), m03 (electrograms), m07 (inverse problems), m12 (surface ECG and the forward problem), m32 (ECG imaging).

References

Brody, D. A. 1956. “A Theoretical Analysis of Intracavitary Blood Mass Influence on the Heart-Lead Relationship.” Circulation Research 4 (6): 731–38. https://doi.org/10.1161/01.res.4.6.731.
Plonsey, R., and D. B. Heppner. 1967. “Considerations of Quasi-Stationarity in Electrophysiological Systems.” The Bulletin of Mathematical Biophysics 29 (4): 657–64. https://doi.org/10.1007/BF02476917.
Plonsey, Robert, and Roger C. Barr. 2007. Bioelectricity: A Quantitative Approach. 3rd ed. Springer. https://doi.org/10.1007/978-0-387-48865-3.
Rudy, Y., and R. Plonsey. 1979. “The Eccentric Spheres Model as the Basis for a Study of the Role of Geometry and Inhomogeneities in Electrocardiography.” IEEE Transactions on Biomedical Engineering 26 (7): 392–99. https://doi.org/10.1109/tbme.1979.326417.